The prom­ise and pit­falls of AI

Financial Nigeria Magazine - - Contents - By Jac­ques Bughin , Nico­las van Zee­broeck

The AI revo­lu­tion will bring short-term pain be­fore long-term gains. If that pain oc­curs against a back­drop of frus­tra­tion with the un­equal dis­tri­bu­tion of AI's ben­e­fits, it may trig­ger a back­lash against tech­nolo­gies that could oth­er­wise pro­duce a vir­tu­ous cy­cle of higher pro­duc­tiv­ity, in­come growth, and em­ploy­ment-boost­ing de­mand.

Like any trans­for­ma­tive trend, the rise of ar­ti­fi­cial in­tel­li­gence (AI) poses both ma­jor op­por­tu­ni­ties and sig­nif­i­cant chal­lenges. But the gravest risks may not be the ones most of­ten dis­cussed.

Ac­cord­ing to new re­search from the McKin­sey Global In­sti­tute (MGI), AI has the po­ten­tial to boost over­all eco­nomic pro­duc­tiv­ity sig­nif­i­cantly. Even ac­count­ing for tran­si­tion costs and com­pe­ti­tion ef­fects, it could add some $13 tril­lion to to­tal out­put by 2030 and boost global GDP by about 1.2% per year. This is com­pa­ra­ble to – or even larger than – the eco­nomic im­pact of past gen­eral-pur­pose tech­nolo­gies, such as steam power dur­ing the 1800s, in­dus­trial man­u­fac­tur­ing in the 1900s, and in­for­ma­tion tech­nol­ogy dur­ing the 2000s.

Per­haps the most dis­cussed con­cern about AI is the prospect that in­tel­li­gent ma­chines will re­place more jobs than they cre­ate. But MGI’s re­search found that the adop­tion of AI may not have a sig­nif­i­cant ef­fect on net em­ploy­ment in the long term. Ex­tra in­vest­ment in the sec­tor could con­trib­ute 5% to em­ploy­ment by 2030, and the ad­di­tional wealth cre­ated could drive up labour de­mand, boost­ing em­ploy­ment by an­other 12%.

But while the over­all pic­ture is pos­i­tive, the news is not all good. For one thing, it is pos­si­ble that it will take time for AI’s ben­e­fits – par­tic­u­larly with re­gard to pro­duc­tiv­ity – to be felt. In­deed, MGI’s re­search sug­gests that AI’s con­tri­bu­tion to growth may be three or more times higher by 2030 than it is over the next five years.

This is in line with the so-called Solow com­puter para­dox: pro­duc­tiv­ity gains lag be­hind tech­no­log­i­cal ad­vances – a no­table phe­nom­e­non dur­ing the dig­i­tal revo­lu­tion. This is partly be­cause, ini­tially, economies face high im­ple­men­ta­tion and tran­si­tion costs, which es­ti­mates of AI’s eco­nomic

im­pact tend to ig­nore. MGI’s sim­u­la­tion sug­gests that these costs will amount to 80% of gross po­ten­tial gains in five years, but will de­cline to one-third of those gains by 2030.

The more trou­bling po­ten­tial fea­ture of the AI revo­lu­tion is that its ben­e­fits are not likely to be shared eq­ui­tably. The re­sult­ing “AI di­vides” will re­in­force the dig­i­tal di­vides that are al­ready fu­elling eco­nomic in­equal­ity and un­der­min­ing com­pe­ti­tion. These di­vides could emerge in three ar­eas.

The first di­vide would emerge at the com­pany level. In­no­va­tive, lead­ing-edge com­pa­nies that fully adopt AI tech­nolo­gies could dou­ble their cash flow be­tween now and 2030 – an out­come that would likely en­tail hir­ing many more work­ers. These com­pa­nies would leave in the dust those that are un­will­ing or un­able to im­ple­ment AI tech­nolo­gies at the same rate. In fact, firms that do not adopt AI at all could ex­pe­ri­ence a 20% de­cline in their cash flow as they lose mar­ket share, putting them un­der pres­sure to shed work­ers.

The se­cond di­vide con­cerns skills. The pro­lif­er­a­tion of AI tech­nolo­gies will shift labour de­mand away from repet­i­tive tasks that can more eas­ily be au­to­mated or out­sourced to plat­forms, to­ward so­cially or cog­ni­tively driven tasks. MGI’s mod­els in­di­cate that job pro­files char­ac­ter­ized by repet­i­tive tasks and lit­tle dig­i­tal knowhow could fall from some 40% of to­tal em­ploy­ment to near 30% by 2030. Mean­while, the share of jobs en­tail­ing non-repet­i­tive ac­tiv­i­ties or re­quir­ing high-level dig­i­tal skills is likely to rise from some 40% to more than 50%.

This shift could con­trib­ute to an in­crease in wage dif­fer­en­tials, with around 13% of the to­tal wage bill po­ten­tially shift­ing to non-repet­i­tive jobs re­quir­ing high-level dig­i­tal skills, as in­comes in those fields rise. Work­ers in the repet­i­tive and low-dig­i­tal-skills cat­e­gories may ex­pe­ri­ence wage stag­na­tion or even re­duc­tion, con­tribut­ing to a de­cline in their share of the to­tal wage bill from 33% to 20%.

The third AI di­vide – among coun­tries – is al­ready ap­par­ent, and seems set to wi­den fur­ther. Those coun­tries, mostly in the de­vel­oped world, that es­tab­lish them­selves as AI lead­ers could cap­ture an ad­di­tional 20-25% in eco­nomic ben­e­fits com­pared with to­day, while emerg­ing economies may ac­crue only an ex­tra 5-15%.

The ad­vanced economies have a clear ad­van­tage in adopt­ing AI, be­cause they are fur­ther along in the im­ple­men­ta­tion of pre­vi­ous dig­i­tal tech­nolo­gies. They also have pow­er­ful in­cen­tives to adopt AI: low pro­duc­tiv­ity growth, ag­ing pop­u­la­tions, and rel­a­tively high labour costs.

By con­trast, many de­vel­op­ing economies have in­suf­fi­cient dig­i­tal in­fra­struc­ture, weak in­no­va­tion and in­vest­ment ca­pac­ity, and thin skills base. Add to that the mo­ti­va­tion-damp­en­ing ef­fects of low wages and am­ple space for pro­duc­tiv­ity catch-up, and it seems un­likely that these economies will keep pace with their ad­vanced coun­ter­parts in AI adop­tion.

The emer­gence or ex­pan­sion of these AI di­vides is not in­evitable. In par­tic­u­lar, de­vel­op­ing economies can choose to take a for­ward-think­ing ap­proach that in­cludes strength­en­ing their dig­i­tal foun­da­tions and ac­tively en­cour­ag­ing AI adop­tion. And, to en­sure that their chang­ing work­place needs are met, firms can take a more ac­tive role in sup­port­ing ed­u­ca­tional up­grad­ing and con­tin­u­ous learn­ing for lower-skill peo­ple.

More­over, these di­vides are not nec­es­sar­ily a neg­a­tive de­vel­op­ment. The re­al­lo­ca­tion of re­sources to­ward high­er­per­form­ing com­pa­nies makes economies health­ier, po­ten­tially pro­vid­ing them with new com­pet­i­tive ad­van­tages vis-à-vis other coun­tries.

But the risks posed by these di­vides should not be un­der­es­ti­mated. Vi­sion and per­se­ver­ance are es­sen­tial to make the AI revo­lu­tion work, be­cause it will bring short­term pain be­fore long-term gains. If that pain oc­curs against a back­drop of frus­tra­tion with the un­equal dis­tri­bu­tion of AI’s ben­e­fits, it may trig­ger a back­lash against tech­nolo­gies that could oth­er­wise pro­duce a vir­tu­ous cy­cle of higher pro­duc­tiv­ity, in­come growth, and em­ploy­ment-boost­ing de­mand.

Jac­ques Bughin is a di­rec­tor of the McKin­sey Global In­sti­tute and a se­nior part­ner at McKin­sey & Com­pany.

Nico­las van Zee­broeck is Pro­fes­sor of In­no­va­tion, IT Strat­egy and Dig­i­tal Busi­ness at Solvay Brus­sels School, Univer­sité li­bre de Brux­elles. Copy­right: Project Syn­di­cate

Work­ers in the repet­i­tive and low-dig­i­tal-skills cat­e­gories may ex­pe­ri­ence wage stag­na­tion or even re­duc­tion, con­tribut­ing to a de­cline in their share of the to­tal wage bill from 33% to 20%.

A cross-sec­tion of in­ductees at the Char­tered In­sti­tute of Bankers of Nige­ria

Newspapers in English

Newspapers from Nigeria

© PressReader. All rights reserved.